12 PROCEEDINGS OF THE AMERICAN ACADEMY. 



one at lower pressures up to 2000 kgm., and the second between 2000 

 kgm. and the maximum, 12000 kgm. A similar method was found 

 necessary in the previous work on water. The reason for this is that 

 at high pressures the moveable piston takes a permanent set, and 

 moves with extreme friction against the sides of the cylinder. The 

 result is that if the pressure has once been pushed to the maximum the 

 piston will not return far enough to reach the low pressures again. 

 The lowest pressure that it was possible to reach after the maximum 

 varied somewhat, but was usually in the neighborhood of 1000 kgm. 

 In order completely to cover the field, therefore, two sets of experi- 

 ments were necessary, one for the high and the other for the low 

 pressures. To make the results comparable, the low pressure meas- 

 urements were carried some distance into the high pressure domain, 

 the upper limit of the low pressure measurements being about 2000 

 kgm. 



The chief variation from the method of the earlier work is in finding 

 the thermal expansion. The method of the earlier work involved 

 two steps. The first was to change the temperature at constant 

 volume. This change of temperature, if it were an increase, produced 

 as a secondary effect a rise of pressure. The second step was to lower 

 the pressure to the original value preparatory to the next change of 

 temperature. The withdrawal of the piston necessary to effect this 

 lowering of pressure was always attended by slight irregularities; 

 apparently the piston does not begin its regular march until after 

 some slight motion has taken place. So that although perfect regu- 

 larity was found in the larger motions of the piston of the compressi- 

 bility determinations, yet the initial irregularity was sufficient to be 

 disturbing with the smaller motions accompanying the temperature 

 changes. The difficulty was avoided by keeping the piston constant 

 in position during a series of temperature changes, and allowing the 

 pressure to vary accordingly. The variation of pressure for a change 

 of temperature of 60°, from 20° to 80°, was about 1000 kgm. at the 

 highest pressures, and proportionally less at the lower pressures. 

 Temperature measurements were made at six values of the average 

 pressure with the apparatus set up for the higher pressures, above 

 2000 kgm., and at two mean pressures with the apparatus set up for 

 the lower pressures, below 2000 kgm. Readings were made at 20° 

 intervals, as in the work on water. At 2000 kgm., for example, the 

 temperature was raised from 20° to 40° and the increase of pressure 

 noted, then to 60° and the increase of pressure noted again, and 

 finally from 60° to 80° and once again the increase of pressure noted. 



